Best Practices: Data on the Web – #2 Provide Descriptive Metadata!

Yesterday, the post about the First Best Practice, which is Provide Metadata, was published. Today, I will discuss the second best practice from the Data on the Web Best Practices document, produced by the DWBP WG.

BP2: Provide descriptive metadata

Provide metadata that describe the overall features of datasets and distributions.

Providing descriptive information about datasets enables user agents to automatically discover datasets available on the Web, as well as allowing humans to understand the nature of the dataset and its distributions.

Intended Outcome

By doing this, humans will be able to interpret the nature of the data in the dataset and its distributions. In addition, software agents will automatically discover the datasets and their distributions.

Data in a WEB publication

Possible Implementation

Descriptive metadata can include the following features of a dataset:

  • The title and description of the dataset
  • Keywords that describe the content
  • The publication date of the dataset
  • The entity responsible for making the data available
  • The contact point for the dataset
  • The geographical coverage of the dataset
  • The temporal period covered by the dataset
  • The themes or categories of a given dataset

Descriptive metadata can include the following features of a distribution:

  • The title and description of the distribution
  • The publication date of the distribution
  • The media type of the distribution

The machine-readable version of the descriptive data can be provided using a W3C-recommended vocabulary specifically designed to describe datasets. For example, the Data Catalog Vocabulary. It provides a framework in which datasets can be described as abstract entities.

See the example below which provides machine-readable data:

It shows how to use the Data Catalog Vocabulary (DCAT) to provide machine-readable data for the bus stops dataset (stops-2015-05-05). The dataset has a .csv distribution (stops-2015-05-05.csv) which is also described using the >Data Catalog Vocabulary (DCAT). The dataset is classified under the domain represented by the relative URI "mobility".
This domain can be defined as part of a set of domains identified by the URI themes. To describe both the concepts and the concept scheme, John used SKOS. To express update frequency, an instance of the “Content-Oriented Guidelines”, developed as part of the W3C “Data Cube” vocabulary, was used. John chose to describe the spatial and temporal coverage of the dataset using URIs from Geonames and the Interval dataset from data.gov.uk, respectively.

:stops-2015-05-05
a dcat:Dataset ;
dct:title "Bus stops of MyCity" ;
dcat:keyword "transport","mobility","bus" ;
dct:issued "2015-05-05"^^xsd:date ;
dcat:contactPoint <http://data.mycity.example.com/transport/contact> ;
dct:temporal <http://reference.data.gov.uk/id/year/2015> ;
dct:spatial <http://www.geonames.org/3399415> ;
dct:publisher :transport-agency-mycity ;
dct:accrualPeriodicity <http://purl.org/linked-data/sdmx/2009/code#freq-A> ;
dcat:theme :mobility ;
dcat:distribution :stops-2015-05-05.csv ;
.
:mobility
a skos:Concept ;
skos:inScheme :themes ;
skos:prefLabel "Mobility"@en ;
skos:prefLabel "Mobilidade"@pt
.

:themes
a skos:ConceptScheme ;
skos:prefLabel "A set of domains to classify documents" ;
.

:stops-2015-05-05.csv
a dcat:Distribution ;
dct:title "CSV distribution of stops-2015-05-05 dataset" ;
dct:description "CSV distribution of the bus stops dataset of MyCity" ;
dcat:mediaType "text/csv" ;
.

Here you can also take a look at an example of descriptive metadata aimed at human understanding.

See? The second best practice in the document is not rocket science either. In fact, the practice of using descriptive metadata promises long-term gains, since the more metadata we have, the richer and more standardized they are, the easier it is to use them for cross-referencing and analysis.